The nature of human serum insulin like activity (ILA): characterization of ILA in serum and serum fractions obtained by acid ethanol extraction and adsorption chromatography

The nature of human serum insulin-like activity

(ILA): characterization of ILA in serum and

serum fractions obtained by acid-ethanol

extraction and adsorption chromatography

Philip L. Poffenbarger, … , Dieter K. Hepp, Robert H.

Williams

J Clin Invest. 1968;47(2):301-320. https://doi.org/10.1172/JCI105726.

Studies were undertaken in an attempt to clarify the apparent heterogeneous nature of human serum insulin-like activity. Methods of preparative zone electrophoresis on Pevikon, acid-ethanol extraction of trichloroacetic acid serum protein precipitates, adsorption

chromatography on DEAE-cellulose and Dowex 50, gel filtration chromatography, and insulin antiserum immunoreactivity were used. The results establish the presence of a substance in serum with in vitro biological properties similar to insuln but with different physicochemical properties. The major portion of serum ILA measured by bioassay

techniques can be attributed to the effects of this substance. Whereas the in vitro biological effects of this substance on muscle and adipose cells were similar to those of crystalline insulin, the substance is distinguished from insulin by: (1) the failure of insulin antiserum to inhibit its in vitro biological effect; (2) a slower electrophoretic mobility (in the gamma-beta globulin zone); and (3) a larger molecular weight, between 40,000 and 50,000 in these studies. It is similar to insulin since both are soluble in acid-ethanol. The results further indicate that previously described insulin-like activity in gamma-beta globulin preparations, the major portion of total serum insulin activity described in acid-ethanol extracts of serum, “bound” insulin, “atypical” insulin, and antibody nonsuppressible insulin-like activity

bioassayed in diluted serum are all one and the same substance. Research Article

Find the latest version:

The

Nature of Human Serum Insulin-Like

Activity (ILA): Characterization of ILA in Serum

and

Serum Fractions Obtained by Acid-Ethanol

Extraction and Adsorption Chromatography

PmuP L. POFEmNBARGER, JoHN W. ENSJNCK, Dmwrmi K. HEPP, and ROBERT H. WILLIAMS

FromtheDivisionofEndocrinology and Metabolism, Department ofMedicine,

University ofWashington SchoolofMedicine,Seattle,Washington

ABSTRACT Studies were undertaken in an attempt to clarify the apparent heterogeneous

na-ture of human serum insulin-like activity. Meth-odsof preparative zone electrophoresis on Pevikon,

acid-ethanol extraction of trichloroacetic acid

se-rum protein precipitates, adsorption chromatog-raphy on DEAE-cellulose and Dowex 50, gel fil-tration chromatography, and insulin antiserum immunoreactivity were used. The results estab-lish the presence of a substance in serum with in vitro biological properties similar to insuln but with different physicochemical properties. The major portion of serum ILA measured by bioassay techniques can be attributed to the effects of this substance. Whereas the in vitro biological effects of this substance onmuscleand adipose cells were similartothose ofcrystalline insulin, the substance is distinguished from insulin by: (1) the failure of insulin antiserum to inhibit its in vitro

biologi-cal effect; (2) a slower electrophoretic mobility

(in the gamma-beta globulin zone); and (3) a larger molecular weight, between 40,000 and

50,-000 in these studies. It is similar to insulin since both are soluble in acid-ethanol. The results

fur-This work was presented in part as an abstract at the 26th Annual Meeting of The American Diabetes As-sociation.

Address requests for reprints to John W. Ensinck,

M. D., Department ofMedicine, Universityof Washington

SchoolofMedicine, Seattle, Wash. 98105.

Received for publication 23 February 1967 and in

re-vised form 14September1967.

ther indicate that previously described insulin-like activity in gamma-beta globulin preparations, the major portion of total serum insulin activity de-scribed in acid-ethanol extracts of serum, "bound"

insulin, "atypical" insulin, and antibody

nonsup-pressible insulin-like activity bioassayed in diluted serum areallone and the same substance.

INTRODUCTION

The term insulin-like activity (ILA) has evolved from the use of insulin bioassay methods which originally were thought to measure the level of

insulin in serum (1, 2). However, discrepancies

in dataobtained from thebioassayofserumbythe rat hemidiaphragm and epididymal fat pad meth-ods suggested that these tissues were responsive to multiple factors influencing glucose uptake and

utilization (3). This concept was substantiated by

the development of the more specific insulin im-munoassay. Normalfasting serumlevels of insulin derived from immunoassay data have

approxi-mated 15-20 uU/ml (14), whereas fasting levels of "insulin" measured invitro bioassay techniques

havevaried between 40

_pU

and 20mU/ml (1, 3).

The nature and role of components of ILA,

other than insulin, have been the subject of much

speculation. This activity has been attributed to

nonspecific effects of serum proteins in vitro

(5),

and furthermore, because serum ILA does not

completely disappear after pancreatectomy and is

notaltered by various physiological and pathologi-cal states, its biologipathologi-cal significance has been ques-tioned (6-8). Nevertheless, variations in serum

ILA have been observed under certain conditions, and a possible physiological role has been im-plied (9, 10). By means of various protein separa-tion procedures, several forms of ILA have been detected. One form has been found to migrate with beta and gamma globulins in serum frac-tionated by cold ethanol precipitation, ammonium

sulfate plus DEAE-cellulose, and zone

electro-phoresis; however, no agreement has been reached regarding the identity of this activity (11-20).

Because crystalline insulin is found with an alpha globulin in most of these procedures (17-21), it has been postulated that ILA in the beta and gamma globulins represents either binding of insulin to protein or, alternatively, other com-pounds with insulin-like effects (17, 18, 21).

Another form of ILA, referred to as total se-rum insulin activity (TSIA), is demonstrable in acid-ethanol extracts of serum (22-24). These extracts have been prepared by the hydrochloric acid-ethanol method for extraction of insulin (25) and the trichloroacetic acid-ethanol procedure for extraction of albumin (26). Because TSIA stimu-lated glycogen synthesis in muscle, it has been con-sidered to represent insulin (22, 27), however, the levels of TSIA ranging from 0.8-7 mU/ml

were inconsistent with the values of immunoas-sayable insulin in these extracts (28).

In addition, by means of the rat epididymal fat

pad,ILA has been demonstrated in unfractionated

serum both in the presence and absence of insu-lin antiserum (9, 29). The activity neutralized by

insulin specific antiserum has been called "typical"

or suppressible ILA, whereas the greater amount of activity not inhibited by antibody has been termed "atypical" or nonsuppressible ILA

(NSILA).

Recently, Bfirgi, Froesch and co-workers have reported that the metabolic action of a partially purified preparation of human

NSILA resembles that of insulin by

several,

pa-rameters tested both in vitro and in vivo (30, 31). Following the observations of Beigelman and as-sociates that serum ILA was no longer detectable after the passage of blood through a cation

ex-change resin (11), it was shown that ILA, re-tained on the resin, could be eluted with acid or

alkaline solutions, which contrasted to the

be-havior of crystalline insulin (32). Based on these findings, Antoniades and colleagues concluded

that the major portion of insulin in serum was in a complex form which they termed "bound" in-sulin (32, 33). Although "bound" insulin has been found to simulate the action of insulin in several biological systems, immunological and

physico-chemical data have failed to corroborate a

struc-tural relationship to insulin (34-37).

Although the exact nature of "bound" insulin, "atypical" insulin, TSIA, and NSILA has not been clarified, several lines of evidence support a direct relationship to one another as

suggested

by Kipnis and Stein

(37)

and Antoniades (10). "Bound" insulin and NSILA have been found to have electrophoretic mobilities in the beta

globu-linregion (30, 38) and theapproximate molecular weight of "bound" insulin, "atypical" insulin, and

NSILA have been reported as

60,000-100,000,

30,000, and

80,000-160,000,

respectively (38, 9, 30). In addition,

physiological

studies in humans

(39) and dogs (40) have suggested that "bound" insulin and "atypical" insulin may have been formed in an extrapancreatic site. Moreover, the in vitro and in vivo metabolic effects of "bound" insulin and a partially

purified preparation

of

NSILA are strikingly similar

(41,

31). However,

the results of other metabolic studies have been divergent. Glucose loading in humans

apparently

initiates a decrease in serum levels of "bound" insulin (42, 10), whereas

"atypical"'

insulin or

NSILA remains

unchanged

(29, 43)

or increases (44). Thus, although considerable data are avail-able to infer that "bound"

insulin, "atypical"

in-sulin,

TSIA,

and NSILA are

probably

similar,

the lack of

comparative

studies and variance in

metabolic data have resulted in caution of such an

interpretation

(31).

Because of the lack of a clear

relationship

among the various forms of

ILA,

the

following

experiments were

performed

to determine some

of the

biological

and

physiochemical

properties of ILA in human serum. ILA was measured in the presence and absence of insulin antiserum

by

means of the isolated rat

_{hemidiaphragm}

and fat cell bioassays. The ILA in serum, acid-ethanol

extracts of serum, beta and gamma

globulin

com-ponents of serum obtained

by

zone

electrophore-sis, and fractions of serum obtained

by

adsorption

im-munoreactivity, electrophoresis, and gel filtration. The results indicate that ILA associated with beta and gamma globulins, most TSIA measured in serum acid-ethanol extracts, "atypical" insu-lin, NSILA, and "bound" insulin are one sub-stance, biologically similar to insulin but differing

in immunoreactivity, electrophoretic mobility, and molecular size.

METHODS

Blood was removed from three fasting, healthy, nonobese donors at one time, the serum separated, pooled, and stored at -20'C. Throughout a 6-8 wkperiod, five such serum pools were obtained and stored separately.

Insulin preparations and antisera. Crystalline bovine insulin1 used in these studies was six times recrystallized with a specific activity of 22.2 U/mg. Insulin-'I and hI

were prepared by a modification of the chloramine-T method of Hunter and Greenwood (45) and degradation

products were removed by cellulose chromatography (46). Oxidized sulfo-B chain of insulin (BSSO3) was obtained from crystalline insulin by the method of Dixon and Wardlaw (47) and radioiodinated by a procedure

similar to that used for insulin.

Antiserum to a mixture of bovine and porcine insulin was developed in guinea pigs by the method of Moloney

and Coval (48). Antisera at dilutions of 1: 100 in the rat

hemidiaphragm procedure and 1: 500 in the fat cell as-say suppressed completely the effect of 1000jAUof insulin per milliliter of incubation medium.

Bioassay procedures. The rat hemidiaphragm bioassay

described by Vallance-Owen and Hurlock was modified as detailed previously (28). Glucose was measured in a Technicon AutoAnalyzer by the method of Hoffman

(49) and the glucose uptake expressed asmicrogramsper

milligram of dry weight during 90 min incubation. Total

diaphragm glycogen, determined on an alkaline digest of the dried muscle (50) by a method using the anthrone

reagent (51), was expressed as micrograms ofglycogen/

milligram of dry muscle during 90 min incubation. Sta-tistical comparison of group means was performed where

appropriate, and SEM was calculated by conventional

techniques (52).

The rat epididymal fat cell bioassay was identical to that reported byGliemann (53). Lyophilized

acid-ethanol-extracted serum albumin and beta globulin prepared by zone electrophoresiswere added to the incubation medium in varying amounts. The final protein concentration was

adjustedto'40mg/ml byvarying thequantityofalbumin.2

Acid-ethanol extraction of serum. Serum was ex-tracted by a modification (28) of the trichloracetic acid

(TCA)-ethanol procedure described by Debro, Tarver, and Korner (26).Although albumin is the major protein

1_{Donated by Boots Pure Drug Company, Ltd.,}

Not-tingham, England.

2_{Fraction V, Lot B 23709 or B 23809, Armour}

Phar-maceutical Co., Kankakee, Ill.

extracted by this procedure (26), other components of

serum have been detected3 (54) and for descriptive con'

venience, the extract will be referred to as "albumin."

The extract in 1% TCA-95%o ethanol was dialyzed in

boiled, grade 27, Visking cellophane casing (28) for 24 hr against runningtap water, 48 hr in six consecutive 10 liters of deionized water at 4VC, and lyophilized. In

some experiments the ILA in these acid-ethanol extracts was partialy separated by isoelectric precipitation of al-bumin. This was accomplished by adjusting the pH of the TCA-ethanol extract to 4.9 with 5 N ammonium

hy-droxide and precipitated albumin was separated by

cen-trifugation. The resulting supernatant was lyophilized

after extensive dialysis as detailed above. The average

yield of supernatant material was 350 mg/100 ml of se-rumextracted.

Anionic exchange chromatography. DEAE-cellulose4

waspreparedfor usebyalternate washing with0.1 N

so-dium hydroxide, 0.1 N phosphoric acid, and equilibration

ofthehydroxide form in 0.001 M phosphatebuffer, pH 8. Serum (20 ml) was applied to a column (52X2 cm)

containing 30 g of cellulose and fractional elution of se-rum protein was performed with 0.001 M phosphate

buf-fer, pH 8 (Fraction I),and 0.1 M phosphate buffer, pH 7

(Fraction II),atflow rates of 100 ml/hr. Fractions I and II were lyophilized after extensive dialysis in boiled, grade 18, Visking cellophane casing against running tap water and deionizedwater.

Serum (20 ml) containing 165,000 cpm of insulin-'I (SA was 250 mc/mg) was subjected to DEAE-cellulose

chromatography under similar conditions as above,

ex-cept after elution ofFractionII, the cellulose was washed with 6M acetic acid (Fraction III). Radioactivity in ali-quots from the seperate fractions was measured in an

autogamma well counter.

Bovine crystalline insulin (25 mg) dissolved in 1.0 ml of 0.1 N HCl, was added to 10 ml of 0.001 M phosphate buffer, pH 8.0, and the final pH was adjusted to 8 with NaOH. The solution was applied to the DEAE-cellulose column and eluted with buffers and acetic acid as de-scribed. Protein in the collected fractions was determined

by absorbency at 280 mgt and the fractions were desalted by dialysis and lyophilized.

Cationic exchange chromatography. Dowex 50X8,5

100-200 mesh, was prepared by the procedure outlined by

Antoniades and Gundersen (55) and was used in the Na+cycle, pH 6.6 ± 0.2. Chromatographywas performed in glass columns (25X3.5 cm) with an amount of wet resin equivalent to serum (v/v). Serum was passed through the resin at flow rates between 3-5 ml/min at room temperature, and the resin was subsequentlywashed with 3 volumes of cold 0.15 M NaCl. The combined se-rum effluent and wash were desalted by dialysis and ly-ophilized. Trichloracetic acid-ethanol extraction of this fraction was performed by dissolving 700 mg of the

ly-3Poffenbarger, P. L., and J. W. Ensinck. Unpublished

observations.

4Cal Biochem, Los Angeles, Calif.

ophilized protein in 10 ml of Gey and Gey buffer at pH 8 (56). Extraction was then performed as described for serum (28).

After passage of serum, the protein retained on the resin was eluted with 2 volumes of 0.02 N NH4OH into an equivalent quantity of 0.2 N sulfuric acid under con-stant stirring. Flow rates were maintained at 15-20 ml/

min and the pH of the collected solution was adjusted to

7.4, and the solution was subsequently lyophilized.

Pro-tein concentration, determined by a modification of the method of Lowry, Rosebrough, Farr, and Randall (57)

varied between 200-300 ,ug/ml, calculated on the basis of equivalence to volume of serum applied to the resin. 24 hr before assay by the isolated rat hemidiaphragm

method, a weighed amount of the lyophilized alkaline eluate, equivalentto the calculated concentration in 10 ml of serum, was dissolved in 5-6 ml of Gey and Gey buf-fer and dialyzed in boiled, grade 18, Visking casing

against 100 volumes of Gey and Gey buffer containing

glucose (3 mg/ml). The retentate was adjusted to a final volume of 10 ml with buffer plus glucose. Prepara-tion of the alkaline eluate for assay in the isolated rat

epididymal fatcell system was performed in similar fash-ion except that Krebs-Ringer bicarbonate buffer with glucose (0.1 mg/ml) was used. Albumin (50 mg/ml) and glucose-1-"4C (0.05 sic/incubation flask) were added after completion of dialysis.

Preparative zone electrophoresis. Electrophoresis was

performed in a block (50.5X37.5X1.5 cm) of Pevikon,6

equilibrated in 0.35 M Tris-EDTA-borate buffer, pH 9

(58). Serum was equilibrated by dialysis in buffer and

applied to the origin (1 X35 cm) located 12 cm from the cathode. In the experiments in which acid-ethanol-extracted "albumin," supernatant material, and alkaline eluate protein were subjected to electrophoresis, serum was run in parallel as a control for division of the block.

Electrophoresis was performed at 4°C for 20-22 hr with a potential gradient of 8 v/cm and 100-120 ma current

flow. Subsequently, the block was cut into 1 cm sections and protein eluted with 30 ml of 0.01 M

Tris-EDTA-borate buffer. The protein content was determined by

absorbence at 280 m/u. The four major protein fractions (gamma, beta, alphaa globulin, and alpha, globulin-albu-min) were segregated, desalted by dialysis, and

lyophi-lized. Average recovery varied from 80 to 92% of the

original protein fractions of serum, determined by

quan-titative electrophoresis on cellulose acetate.

Gel filtration. Ascending gel filtration was performed

at room temperature through a column (180 X 2 cm) of

Sephadex G-1007 in 0.15 M phosphate buffer, pH 7.4, at flow ratesvarying from 25 to 30 ml/hr. The column was

calibrated using the following compounds of known

molecularweight: Dextran20007 (2 million, average mol

wt); human crystalline albumin8 (69,000 mol wt); bo-vine ovalbumin7 (44,000 mol wt); bovine myoglobin9

6_{Superfostat Bolaget, Stockholm,} Sweden.

7Pharmacia Fine Chemicals, Inc., Piscataway, N. J.

8_{Pentex Company, Philadelphia, Penna.}

9Mann Research Laboratories, New York.

(16,900 mol wt); bovinecrystalline insulin10 _(12,000 mol wt); and radioiodinated sulfo-B chain of insulin (4000

mol wt, specific activity 200 mc/mg). 20 mg of each pro-tein and 2X106 cpm of BSS08-'I were dissolved in 5

nil of 0.15 M phosphate buffer, pH 7.4, and separately

filtered. Protein content of each fraction (6ml/tube) was

determinedby the absorbence at 280m/u andradioactivity

was measured in an autogamma well counter.

Distribu-tion coefficients (Kd) were determined for each protein according to the method outlined by Grannath and Flodin

(59).

Protein samples from zone electrophoresis and acid-ethanol extraction of serum were applied separately to the column in 5 ml of phosphate buffer. Alkaline eluate salt (120 mg/5 ml of water) was equilibrated in buffer beforefiltration. Serum (20 ml) was equilibratedin

buf-fer by dialysis and lyophilized.The powder was dissolved

in 5 ml of water andapplied to the column. After deter-mination ofabsorbence at 280 m, in the eluted fractions, the peaks weresegregated into fivefractions, desalted by dialysis, andlyophilized.

RESULTS

ILA in serum

fractionated

byzone electrophore-sis. Resolution of serum protein into four frac-tions was obtained by electrophoresis of 20 ml of

serum from each of four separate serum pools on

Pevikon in 0.35 M Tris-EDTA-borate buffer at pH 9 (Fig. 1). Although adistinct

_alpha,

globulin

was not observed, albumin preparations from the Pevikon blockrevealed the

_alpha,

migrating

com-ponent on cellulose acetate

electrophoresis

(not shown). Insulin-125I added to serum

migrated

in the

_alpha,

globulin-albumin zone. As shown in Figure 1, significant stimulationof glucose uptake

in the rat hemidiaphragm was only observedwith fractions from the beta and gamma

globulin

re-gions with percentage increases in uptake of 80 and

48%o,

respectively.

The effect of guinea pig anti-insulin serum on the stimulation ofglucose uptake

produced by

the gamma and beta globulins from three separate

se-rum pools is shown in Fig. 2. Excess antiserum

(dilution 1:100) wasadded before incubation and

neutralized the effect of 1000

,uU

of insulin per milliliter. However, the

gamma-beta globulin

preparations at protein concentrations of 10 mg/

ml contained ILA, which onbioassay was

equiva-lent to

100-500

jmU of insulin, yet the ILA was

not suppressed by the addition of antiserum. The

inabilityto detect any ILA in the

alpha1

globulin-albumin region where crystalline insulin migrates

10_{Sigma Chemical Company,} St. Louis, Mo.

4t

Origin

_I

_y

_I

J a q2 'l1tAlbumin

qzj

-0.8

-0.6

-16 a

-OAN

-112 t

-0.2 8

4

cm 5 10 15 20 25

FIGURE 1 Electrophoretic mobility of serum ILA and insulin-"'I.

Serum (20 ml) was submitted to electrophoresis on Pevikon in

0.35 MTris-EDTA-borate buffer, pH 9. The resulting protein

frac-tionswerecombined into four pools (+ the origin) as shownbythe

short vertical lines. Each pool was assayed by the rat hemidia-phragm response. Mean net glucose uptake of four experiments (four hemidiaphragms per experimental group) is represented by

the bars. Vertical lines indicate SEM. Gamma, beta, and alpha2

globulins were assayed at 12.5 mg/ml, alpha1 globulin-albumin at

40 mg/ml. The electrophoretic migration of insulin-1I, added to

20 ml of serum (1000 cpm/ml) inone experiment, is shown as the

dottedarea.

(21) may be explained by the levels of insulin

in fasting serum, measured immunologically

(10-20

JfU/ml)

(1, 4), whicharebelow the sensitivity of the hemidiaphragm assay. Thus, these studies demonstrated that a substance with an electro-phoretic mobilitycorrespondingtobetaandgamma globulins was capable, like insulin, of stimulating

glucose uptake in muscle. However, this sub-stance(s) differs from insulin in its

electropho-retic mobility and its inability to be neutralized byinsulin specific antiserum.

ILA in acid-ethanol extracts of serum. The

stimulation of glucose uptake into rat

hemi-|

32-a 16

8

)hragm produced by TCA-ethanol extracts of

m (four separate pools) is illustrated in Fig. t is evident that ILA was extracted into 1%o

A-95%o ethanol with serum albumin and that

ILA was not inhibited by the addition of

in-n specific antiserum. In order to characterize

:her the ILA in the "albumin" preparations, procedure of isoelectric precipitation was

ap-d. After isoelectric precipitation of albumin in A-ethanol serum extracts, the supernatant was

nd to contain mostofthe ILAand no

suppres-ofactivity wasobserved in thepresence of

in-n antibody (Fig. 3). Although complete

sepa-FIGURE2 Effect of insulin antiserum on ILA

of serum proteins prepared by zone

electro-phoresis. Protein fractions were assayed by the

rat hemidiaphragm response in the presence and absence of insulin antiserum (1: 100 dilution). Concentrationofgamma,beta,andalphas globu-lin was 10 mg/ml; alpha,globulin-albumin was

40 mg/ml. Insulin concentration was 1000 AU/

ml. Bars represent the meannet glucose uptake of three experiments (four hemidiaphragms per experimental group) and vertical lines indicate

SEM.

N:4

I I

N 3 Insulin

_ontiserum

added 0

FIGURE 3 Stimulation of glucose uptake into

the rathemidiaphragm by "albumin" and

super-natant material. For preparation of "albumin" and supernatant material, refer to Methods.

Insulin antiserum dilution was 1: 100. Bars

rep-resent mean net glucose uptake in four experi-ments (three hemidiaphragms per experimental

group) and vertical lines indicate SEM.

ration of ILA from albumin was not achieved by this procedure, a two- to three-fold purification

wasaccomplished. Thus,asin thebetaandgamma

globulin fractions from zone electrophoresis of serum, ILA that was not suppressed by insulin

antiserum was also extracted into acid-ethanol.

DEAF-cellulose _{chromatography} of unmodified

serumILA andacid-ethanol-extractedserumILA.

In Table I are shown the effects, on glucose

up-take in isolated rat muscle, of serum (Experi-ment 1), "albumin" (Experiment 2), and super-natant material (Experiment 3) fractionated on

DEAE-cellulose. Fractions I and II for each of the separate experiments were tested with and

without insulin specific antiserum.

In Experiment I, in whichserumwas

chromato-TABLE I

DEAE-Cellulose Chromatography of ILA in Serum and Acid-Ethanol Extracts of Serum

Chromat- Protein GlucoseUptake+

ographic

concen-Experiment Experimentalgroup fraction* tration Fraction alone Antiserum adde(la mg/mI pg/mgdrywt

1 Buffer 20.98 :1 1.46

Insulinll 48.64 i 1.70 21.72 4 2.42

Serum I 10 23.784 2.84 23.90 1.18¶ II 40 29.58 ± 1.18 29.464 1.12**

2 Buffer 20.12 4 0.58

Insulin 47.66 + 1.46 22.32 =1 1.78

"Albumin"Tt I 0.25 18.584 1.16 19.12 i 1.561

II 40 26.34 ± 2.08 26.08 4l 1.16** 3 Buffer 16.50:t0.64

Insulin 44.10 ± 2.30 18.18A 2.12

Supernatantmaterials I 0.10 16.764- 1.82 18.144 _1.381

II 2 38.944 _{1.08 38.68 ± 0.90§§}

* Fraction Iwaseluted with0.001M_{phosphate buffer, pH}8; Fraction II with0.1Mphosphate buffer, pH7. t Each value represents themeanuptake of three hemidiaphragms 41 SD.

§Insulinantiserumdilution is1:100.

Insulinconcentrationwas 1000IAU/mlinthe threeexperiments.

¶ Valuesare notsignificantlydifferent( >0.05)fromuptakein buffer alone.

** Difference between uptake of fraction +antiserumand that inbufferalone issignificantatP <0.02.

ft RefertoMethods forpreparationof "albumin" andsupernatantmaterials.

§§Differencebetweenuptakeof fraction +antiserum-andthat in buffer alone issignificantatP < 0.001.

}$-24-q

q%

Ctt

%t6

thI6 *Z

graphed, Fraction I consisted of gamma globulins and Fraction II contained albumin, alpha and

beta globulins as determined by cellulose acetate

electrophoresis. In Experiments 2 and 3, inwhich 1.5 g "albumin" and 80 mg of supernatant material from acid-ethanol extracts of serum were sub-jected to chromatography, only 5 and 2 mg of protein were recovered, respectively, in Fraction I, whereas Fraction II contained the major amount of protein.

As is apparent from Table I, a significant

in-crease in glucose uptake above that in buffer alone was observed only with the protein in Fraction II from each of the three experimental groups (P < 0.02, < 0.02, and <0.001, respectively). In addi-tion, the ILA in these fractions was not

neutral-ized byinsulin specific antiserum. The lowamounts

of ILA detected in Fraction II in Experiments 1

and 2, in which the protein concentrations were 40 mg/ml, may be due to the presence of an

in-hibitor associated with albumin as described previ-ously (28). Thus, nonsuppressible insulin-like

ac-tivity in serum, in acid-ethanol extracts of serum ("albumin"), and in its concentrate in the super-natant fraction was eluted from DEAE cellulose under identical conditions. In contrast to the

be-havior ofserum ILA,

95%

ofinsulin-'31Iin serum and virtually all crystalline insulin was retained on DEAE-cellulose and approximately

50%

of

both was eluted with 6 N acetic acid (Table II).

The experimental yields of ILA and insulin from

DEAE-cellulose chromatography were below the expected values and probably reflect losses from

TABLE I I

DEAE- Cellulose Chromatography of Crystalline Insulin and Insulin-13sI*

Chromatographic

fraction Insulin-131I Insulin

cpm mg

I 500 0

II 7,700 0

III 71,300 14 Per cent recovery 48 56 * Insulin-'III (165,000 cpm) in 20 ml of serum and 25 mg of crystalline insulin were chromatographed on DEAE-cellulose. Fraction I was--eluted with 0.001 M phosphate buffer, pH 8; Fraction II with 0.1 M phosphate buffer, pH 7; and Fraction III with 6 N acetic acid.

adsorption to cellulose and from processing of the

effluent fractions for assay.

Dowex 50 chromatography of serum ILA. In Table III are shown the results of studies

de-signed to determine the chromatographic behavior

of ILA onDowex 50 X 8 (100-200 mesh).Equal

volumes of pooled human serum were passed

through varying volumesofDowex50equilibrated

in0.15 M NaCl. 99%o of the total protein was re-covered in the NaC1 wash. This protein fraction was assayed directly at 10 mg and 70 mg/ml. In addition, this fraction was extracted with TCA-ethanol and the "albumin" assayed at 10 mg/ml. The material retained on the resin was eluted with

alkali (alkaline eluate) and assayed at concentra-tionsapproximatingthose in theoriginal serum. In

contrast tothe studies of Antoniades and

Gunder-TABLE I II

Dowex 50 Chromatography of Serum ILA*

Glucoseuptakeabovebasalt

Resin Serum Acid-ethanol extract

volume volume NaClwash§ ofNaClwash Alkalineeluatell

ml ml 10mg/ml P 70mg/ml P 10mg/ml P 0.25mg/ml P

10 20 7A0O 0.58 < 0.05 8.52 ± 1.28 <0.02 5.66 =1 1.14 < 0.05 13.56 :1 3.00 <0.001

20 20 2.08 i 0.36 <0.05 5.32 i 0.58 <0.05 1.94 i 1.28 NS 22.58 ± 3.34 < 0.001

30 20 1.20 ± 1.20 NS 4.78 i 1.68 <0.05 -0.42 +0.54 NS 23.90 41 2.58 < 0.001

40 20 2.48 ± 0.76 NS 4.52 ± 0.92 <0.05 1.56± 1.46 NS 25.32 ± 2.04 < 0.001 *_{Chromatography}ofserumwasperformed throughvaryingvolumes of resin.

Glucoseuptakeisexpressedasgg/mgofdrymuscle andrepresentsthemeannetuptakeof three hemidiaphragms =1 SDupvaluesindicate levels ofsignificanceof theuptakein buffer alone(basal).NS refersto p > .05.

§NaClwash refersto serumproteins passed throughthe resin combined with the wash in 0.15 M NaCl.

TABLE IV

Reproducibility of Dowex 50Chromatographic Separation of ILA in Serum*

Insulin Alkalineeluate4

Deter-Glucoseuptake mina- Glucose uptake Specific

above basal tion above basal activity

Mug pro- ;MU ILA/mg

MU/ml tcin/ml protein

100 6.62 + 1.12 1 250 18.104 3.02 2000 200 12.06 i 1.12 2 260 19.844 2.40 2300

400 16.68 ±2.50 3 265 18.62 A 1.82 2000

600 19.30 4 3.40 4 260 18.204 2.24 2000 800 21.24 i 2.52

* Dowex 50 X-8 chromatography of a pool of serum was performed at a resin to serum volume ratio of 1:1 on four separate occasions. All fractionswereassayed by therathemidiaphragm response simultaneously with the insulin dose responsecurve. Glucoseuptake is expressedasjtg/mgofdry muscle and represents the mean net uptake of three hemidiaphragms 4 SD.

IAlkaline eluate refersto the fractioneluted from the resin with 0.02 N NH4OH. The protein concentration of this fraction isequivalenttotheamountof protein eluted per milliliter of serum applied.

sen (60), the bioassay of the alkaline eluate by

the rat hemidiaphragm procedure did not include the presence of adipose tissue extracts. As indi-cated in Table III, ILA was detected in the alka-line eluate from the Dowex column and maximum retention of ILA on the resin was observed at

resin:serum volume ratios of 1: 1 or greater. At

aresin:serumvolume ratio of 1:2, 43%o less ILA was found in the alkaline eluate and

correspond-ingly more ILA was recovered in the serum pro-teins which were not retained on the resin (NaCl wash). It is also apparent that, after acid-ethanol extraction of the serumproteinsin the NaClwash,

additional ILA was not demonstrated. These data, therefore, are compatible with the conclusion that most of the serum ILA is retained on thecationic exchange resin and that ILA is not due to a

non-specific effect of acid-ethanol on serum proteins. In Table IV is shown the ILA in the alkaline eluate from four replicate experiments usingpooled serum at resin: serum volume ratios of 1: 1. It is evident that, within the limits of precision of the

rat

hemidiaphragm procedure,

therecoveryof ILA in the alkaline eluate from Dowex 50X 8 is

rea-sonably reproducible. In contrast, considerable variation in ILA was encountered in alkaline

TABLE V

Variation of AlkalineEluateILA in Sera from Different Human Subjects*

Insulin Alkalineeluatet

Glucose uptake Glucoseuptake Specific

abovebasal Lot above basal activity

ug pro- MUILA/mg

MU/mI cin/ml protein

200 6.98i1:2.66 1 163 16.10 + 1.70 3060

400 14.30+0.80 2 275 12.46 i 1.72 1320

600 17.82 1 1.52 3 294 6.304 3.52 610

800 22.54 +2.80 4 250 13.46 1.90 1560

1000 24.54 1.40 5 430 18.444 1.64 1418 * Dowex 50 X -8 chromatography of serum from five separate lots was performed at a resin to serum volume ratio of1:1.Allfractionswereassayed bytherathemidiaphragmresponsesimultaneouslywiththe insulin dose response

curve.Glucose uptake isexpressedasinTable IV.

TABLE VI

EffectofInsulin AntiserumonBiological Activity ofSerumILASeparated by

Dowex50_{Chromatography}

Glucose_uptake+ Protein

Fraction* concentration Fraction alone P Fractionplus antiserum P

mg/mI jig/mg dryws

Buffer (basal) 22.18 :1 2.20

Insulin 49.20A 3.26 <0.001 24.18 d 1.84 NS

NaClwash 70 28.44 : 1.98 < 0.05 24.564 3.62 NS Alkaline eluate 0.20 37.06 + 3.58 < 0.001 39.14A 0.82 <0.001

*_{For description ofNaCl wash and alkaline eluate fractions, refer to TableIII.Insulin concentration is 1000}

_jU/ml

and insulin antiserum dilution is 1:100.

tEach value representsthemean glucoseuptakeofthree hemidiaphragms:1 SD. ThePvaluesrefertothe level of significance of difference between glucose uptake inbufferalone and that in the corresponding fractions.

eluates obtained from five individual lots of serum (Table V). The protein concentration varied be-tween 163 and 430

_pg/ml

and the specific activi-ties between 610 and 3060 MAU equivalents of in-sulin per milligram of eluate protein. The average of five determinations was 1600

Pu/mg

ofprotein

which is a two- to three-fold greater specific ac-tivity than that reported previously in assay sys-tems employing adipose tissue extracts (34, 41,

60).

Table VI shows the effect on glucose uptake in muscle of serum proteins in the

NaCi

wash and in the alkaline eluate in the presence ofinsulin

antiserum. A small but significant (P< 0.05) in-crease in glucose uptake was observed with the

proteins of the NaCl wash. After the addition of

antiserum,the glucose uptakewas notsignificantly different from the uptake under- basal conditions

or in the NaClfraction without antiserum.

There-fore, although it is difficult to be certain that the ILA was suppressed, it does indicate that little if any nonsuppressible ILA was present in theNaCl wash. In contrast, the large amount of ILA in the alkaline eluate was not neutralized by insulin spe-cific antiserum, and by means of radioimmunoas-say, no immunologically measurable insulin could bedetected.

Dowex 50 chromatography of the ILA in

"al-bumin" preparations. "Albumin" prepared by

TCA-ethanol extraction of serum was passed through Dowex 50 x- 8 under conditions de-scribed for serum. In Table VII is illustrated the effect on glucose uptake in muscle of the proteins intheNaCl wash, and in the alkaline eluate in the presence and absence of insulin antiserum. The ILAin "albumin" that was not neutralized by in-sulinspecific antiserum was retained on the resin, which coincided with the behavior of ILA from

TABLE VI I

Effectof Insulin Antiserum onBiologwa Activity of "Albumin"-ILA Separated by Dowex 50 _{Ohromatography}

Glucoseuptake+

Protein

Fraction* concentration Fraction alone P Fraction plus antiserum P

mg/mI jpg/mgdry wt

Buffer (basal) 22.184 2.20

Insulin 49.204 3.26 < 0.001 24.18 1 1.84 NS

NaCl wash 10 24.88 + 2-.08 NS 24.264 2.44 NS Alkaline eluate 0.32 32.58 41 186 <0.001 31.924 2.64 < 0.001

*_{"Albumin" (1 g in 20 ml ofGeyandGey buffer, pH 8)waschromatographedon Dowex 50 X -8 (as described}

for serum).FordescriptionofNaClwashand alkaline eluatefractions, refer toTable III. Insulin concentration is 1000

pU/ml.

tThe dataare_presentedasinTableVI.

serum subjected to Dowex chromatography (Ta-ble III). Thus, it seems reasonable to conclude the acid-ethanol extraction of serum did not alter the property of this insulin-like substance to ad-sorb tothe Dowex 50 resin.

Dose response relationships of ILA in various protein fractions. Fig. 4 and 5 show the dose

k

941

*:

'Nk

C-4

200400600 800 0.5 1.0 1.5 iLU/ml mg/ml

zz

(4 N1

NZ

0

N.

d4

01

It

VN

tzN

'44

Nk,

FIGURE 4 The effect of varying concentrations of acid-ethanol extracts of serum, beta globulin, and alkaline eluate on glucose uptake and glycogen synthesis in rat

hemidiaphragm. Forpreparationof"albumin," supernatant

material, beta globulin, and alkaline eluate refer to

Methods. The points for glucose (0-0) and glycogen (0-0) represent means of three experiments for the

acid-ethanol extracts,twoexperiments for thebeta

globu-lin fraction, and two experiments for the alkaline eluate protein from separate serum pools. Vertical lines

indi-cate SEM. Insulin dose response curves, which are shown

for comparison, were performed at the same time.

O 60 80 1oO1000 02 04 .06 .08 6

twU/ml mg/ml

FIGURE 5 The effect of varying concentrations of acid-ethanol extracts of serum, beta globulin, and alkaline

eluate on glucose 1-14C oxidation in isolated fat cells. For preparation of "albumin," beta globulin, and alkaline eluate refer to Methods. The various fractions were

as-sayed in the absence (0-0) and presence (0-*) of insulin antiserum (G-PAIS) at 1:500 dilution. Each

point represents the mean of duplicate determinations for the "albumin" and beta globulins and triplicate

de-terminations for the alkaline eluate protein with SD indi-cated by vertical lines. Each preparation is shown with the corresponding insulin dose response.

response curves for glucose uptake and glycogen synthesis in the rat hemidiaphragm and glucose

oxidation in the adipose cell assay for the differ-ent protein fractions containing ILA in the

pres-ence of insulin antiserum. The effect ofcrystalline

insulin is included for comparison. It is evident that the ILA in all of the protein fractions not

only causedglucose uptake in muscle but also

insu-uin by both of these parameters (Fig. 4) ap-proached a maximum at a concentration between 800 and 1000

juU/ml,

whereas with the exception

of the ILA in the alkaline eluate, the increase of glucose uptake and glycogen synthesis by these fractions did not achieve the maximum effect of insulin. The greatest effect (Fig. 4) occurred with beta globulin at 10 mg/ml, supernatant material at 4 mg/ml, and "albumin" at 10 mg/ml. Although beta globulin and supernatant material at these concentrations approximated the calculated

con-centration in serum, the maximum effect of albu-min was seen at one-fourth of its serum

con-centration. With increasing concentration, less apparent ILA was detected, which suggests that inhibition of the ILA had occurred at higher

con-centrations of ILA in fractionated serum by the hemidiaphragm bioassay.

In contrast to the evidence of inhibition of ILA with the above fractions, the ILA in the alkaline

eluate protein didnot appear to be inhibited since it achieved the maximumeffect observed with 800

,AU of insulin. That this lack of inhibition is not

Serum I,

A '- 8

Origin f '4' _-04

Lo' _/ 0_{02 albumin"}

~-Fraction

I--o*-.Fraction

1I*-"Albumin" 08

Supernatant

_16-

60C j>_-1'0'!08408 Serum

D 02

Origin%

a2 aalbumin%%,

Alkaline eluate

the result of lesser amounts of protein in the alka-line eluate is indicated by the observations that purified albumin at concentrations of 50 mg/ml does not inhibit the effect of insulin in the hemi-diaphragm bioassay (28).

Fig. 5 shows the dose response relationships for ILA in "albumin" beta globulin and alkaline eluate protein as assayed in a fat cell bioassay system with and without insulin antiserum. Be-cause of the increased sensitivity of this assay, lower concentrations of the protein fractions were possible. From the curvilinear dose response rela-tionships for the ILA in "albumin" and the alka-line eluate, which were similar in contour to those of insulin, it is concluded that no inhibition was present, whereas the ILA in the beta globulin pro-tein tended to plateau below the insulin response

suggesting the presence of an inhibitor in this fraction. Despite the differences in dose response relationships on muscle and adipose tissue, the ILA in these derivatives of serum exerted a

qualitatively similar biological effect in both assay systems.

FIGURE 6 Electrophoretic mobility of ILA in

"al-bumin," supernatant material, and alkaline eluate.

"Albumin" (2 g/20 ml), supernatant material (0.5

g/20 ml), and alkaline eluate (8 mg) were sub-mitted to zone electrophoresis under conditions de-scribed for serum (Fig. 1) and are shown as B, C,

and E, respectively. In A, electrophoresis of serum (2 ml) in parallel with either "albumin" or

super-%:> natant material served as a marker for

segmenta-'sN tion of the Pevikon block. In B, all of the electro-phoretic fractions were combined into two pools, called Fractions I and II, asindicated by the short

vertical lines. Fractions I and II were assayed by the rathemidiaphragmresponse at10 and 40mg/ml,

respectively. In C, all of the electrophoretic fractions were combined into four pools, as indicatedby the short vertical lines. The four fraction pools were

assayed at 5 mg/ml for the alphas and 20 mg/ml

for thealpha. globulin-albuminfraction pool. Bars represent the mean net glucose uptake for two

experimentsfrom separate serum pools andvertical

lines indicate SEM. In D is shown the migration

of serum proteins run in parallel with the alkaline

eluate protein. In E, the electrophoretic fractions

were combined into five pools, as indicated by the short vertical lines. The bars indicate the mean

glucose uptake ± SD (three hemidiaphragms per

determination) of the five fraction pools tested.

10 15

Electrophoretic mobility of ILA in "albumin,"

supernatant material and alkaline eluate protein.

The migration of ILA in "albumin," supernatant

material and alkaline eluate protein was

deter-mined by zone electrophoresis in

Tris-EDTA-borate buffer, pH 9, with serum run in parallel as reference. As illustrated in Fig. 6, the ILA in

all three serum protein fractions was distributed

in zones corresponding to gamma and beta

globu-lins. Pevikon electrophoresis of "albumin" and

supernatantmaterial resultedinseparation ofILA from the major protein of these preparations. The heterogeneity of these extracted products is

sug-gested by the presence ofan

alpha,

migrating

pro-tein during electrophoresis ofboth "albumin" and

supernatantmaterial; whether this secondarypeak

consisted of albumin with slower migration or an

alpha globulin was not ascertained. The

electro-phoretic mobility ofthe ILA inthe alkaline eluate

corroborates the observations of Antoniades,

Hu-ber, and Boshell (38). Because of the small

amounts of protein used in the zone

electrophore-sis, no protein peaks could be measured from the

block, however subsequent to electrophoresis of the alkaline eluate on cellulose acetate in 0.075 M

barbital buffer, pH 8.6, protein was only detected

in the alpha1 globulin region. Thus from the

mi-gration of the ILA in the gamma-beta zone and

the evidence that most of the protein was located

in the

_alpha,

globulin region, it would appear

likely that higher specific activity for the ILA could be achieved.

Estimation of molecular weight of ILA by gel

filtration. Bymeans of gelfiltration with Sephadex

G-100 equilibrated in 0.15 M phosphate buffer,

pH 7.4, the approximate molecular size of the ILA in unfractionated serum, gamma and beta

globulin preparations, acid-ethanol extracts of

serum, and alkaline eluate was determined. As

il-lustrated in Fig. 7, proteins of known molecular

weight were filtered to determine the volume of distribution from which distribution coefficients

(Kd) were calculated.

Gelfiltration of20 ml of serum, concentrated to

5 ml before application to the column, resulted in

an incomplete separation of two heterogeneous protein groups (Fig. 7): The first peak (Kd= 0),

characterized by cellulose acetate electrophoresis,

contained gamma and betaglobulins andan

_alpha,

migrating protein. The second peak (Kd= 0.194)

;4t

".-," 16 ,

:

12-9j3b~ 8.

16

8

16-

2-". 8:

16-

8-

~t1.6-

Z12- 8- 4-i 161

%-12-

8-

a4-

16-121

8-4

l0(D 200 300 400 500

Effluent,ml

0.3

0.1

1.2

0.4 2.0

1.0

-2.0 -1.0 N -0.4

-0.2

-0.8 -0.4

[0.2

-0.1

FIGURE 7 Distribution of ILA by gel filtration of se-rum, gamma, and beta globulin, acid-ethanol extracts of

serum, and alkaline eluate protein. Ascending filtration

was performed through a column (180 cmX2 cm) of Sephadex G-100 in 0.15 M phosphate buffer, pH 7.4.

Proteins of known molecular weight were separately

filtered for standardization. Fractions from each gel fil-tration were arbitrarily combined into fraction pools as

indicated by the short vertical lines. The fraction pools

were assayed by the rat hemidiaphragm response. Bars

represent mean net glucose uptake of each fraction pool for two experiments (three hemidiaphragms per

ex-perimental group) using separate serum pools, and

ver-tical lines indicate SEM.

contained albumin and alpha and beta globulins. The smaller third peak (Kd= 0.924) was not

characterized by electrophoresis. On bioassay of the filtration fractions of serum, the major peak

of ILA was distributed between 200 and 300 ml

of effluent, whereas crystalline insulin (Kd=

0.722) was eluted between 360 and 420 ml. By

means of the immunoassay for insulin by the

Albumin 690000 _{Standorazotion} 0Oolbumin44,000 03

Myoglobinl6b00l3BSS03A'I4,000_ED

Insulin 12,000El

./,

AA

A

Serum

!

'.1.

#-Globulin

Y-Globulin

.\~~~~~~~~~~1

"Albumin"

Supernotont

AlkalineEluote

double antibody technique (62),18,uU of immuno-reactive insulin (IRI) was detectedinthe effluent fraction between 300-400 ml, which corresponded to the distribution volume of crystalline insulin. IRI was not measurable in any of the other fractions.

Beta and gamma globulin preparations from zone electrophoresis of serumwere also separately filtered. A biphasic protein distribution was ob-tained from the beta globulin filtration and pos-sibly correspondedto high molecular weight

glob-ulins (greater than 100,000) and metal-binding

globulins approximating 90,000 molecular weight

(Kd

=0.096) (63). Filtration of gamma globulin

produced a single peak excluded with the void volume andcorrelated with the generally accepted

molecular weight of 160,000 for these proteins (63).

As shown in Fig. 7, the ILA in both the beta and gamma globulin fractions, like that in unfrac-tionated serum, wasdistributed in an effluent

vol-ume of 230 to300 ml, andpartial separation from the major protein peaks was obtained. Similarly,

the ILA in the acid-ethanol-extracted "albumin" and supernatant material was partially separated

from albumin and distributed in the same effluent volume as ILA in unfractionated serum and the beta and gamma globulins from zone

electropho-resis of serum.

Following gel filtration of alkaline eluate, two major protein peaks were observed. The first was

excluded from the gel in the void volume corre-sponding to a substance of molecular weight of

100,000 or greater. The second peak was eluted with a coefficient of distribution (Kd= 0.946) of

a substance of low molecular weight (less than

10,000). The major peak of ILA, assayed by the rat hemidiaphragm, was distributed in the effluent fraction between 200 and 300 ml corresponding to thebehavior of the ILA in the other fractions.

In these studies, the eluates from the gel filtra-tion column were pooled into five fracfiltra-tions, thus somewhat limiting the exact resolution of the elution pattern of ILA; however, since all of the fractions tested encompassed the elution behavior of compounds of molecular weight from above 100,000 to below 4000, it seems reasonable to conclude that most of the biological activity was distributed in a volume of 200-300 ml. By corre-lation with the distributions of the standard pro-teins, the

K.5

for the ILA in all of the preparations

subjected to gel filtration probably ranged from 0.193 to 0.330. A mean Kd of 0.261 for the ILA in serum and fractionated products of serum agrees closely with the Kd of 0.266 for ovalbumin which has a molecular weight approximating 44,000, and therefore it seems likely that the mo-lecular weight of ILA was in the general range of

40,000-50,000.

Effect of temperature and pHon

alkaline

eluate

ILA. As shown in Table VIII, heating insulin

or alkaline eluate ILA at

600

or 80'C for 5 min did not significantly alter the biological activity assayed by the rat hemidiaphragm. In Fig. 8 is illustrated the effect of pH on the biological ac-tivity of alkaline eluate ILA. It is apparent that the ILAwasnotaffected by exposure to pH rang-ing from 2 to 10, however, after incubation by

pH 12,virtuallyall biological activity disappeared. These findings are in agreement with those

re-ported by Burgi, Muller, Humbel, Labhart, and Froesch (30) who observed that a partially

puri-TABLE VI II

EffectofHeatontheBiologicalActivitiesofInsulinand AlkalineEluateILA

Glucose uptake:

pzg/mgdrywt

Experimental*group Concentration Control 5 min 5 min

230C 600C 800C

Buffer 17.30 4 2.00 15.86 d 1.64 16.88 1(0.15

Insulin 1000pU/ml 43.46 4 1.76 43.20+0.50 38.34 ± 1.34

Alkaline eluateprotein 310pg/ml 36.56 + 1.80 36.74 ± 1.36 35.68 ± 0.98

*_{Experimental groupswereprepared inGey and Geybuffer with glucoseand heated in awater bath. Bioassay}

32-Insuiin Cont H

₂₁

0

_I

PH6 p118 PHIO pH12

BaU/ml|

Alkaline Eluole 160ILe protein/ml

FIGURE 8 Effect of pH on biological activity of alkaline eluate ILA.Alkalineeluate (200 mg of salt), dissolved in 20 ml of deionized water, was divided into seven

equal aliquots. After adjustment of pH with sulfuric acid or ammonium hydroxide, the solutions were left at 40C for 4 hr. Dialysis was then performed against

100 volumes of Gey and Gey buffer with glucose added at 4VC in preparation for hemidiaphragm bioassay. The bars represent mean glucose uptake of three

hemidia-phragms ±SD.

fiedpreparation of nonsuppressible ILA was both

heat stable and denatured atalkalinepH.

DISCUSSION

In these studies, fractionation of serum protein was performed by (1) preparative zone electro-phoresis, (2) acid-ethanol extraction of TCA pro-tein precipitates, (3) isoelectric precipitation of

albumin from acid-ethanol extracts, and (4) gel

filtration. Analysis of the resulting serum protein

fractions for ILA demonstrated the presence of

a substance in serum with biological similarities

tocrystalline insulin when tested in vitro, but

dif-ferring from insulin inphysicaland immunological

properties.

Itisestablished thatcrystalline insulin migrates similarly to

_alpha,

globulins

of serum in

electro-phoretic systems that utilize nonadsorbing media at pH values ranging from 7.4to 8.6

(21)

and in

the present studies employing Pevikon at pH 9,

this electrophoretic mobility for insulin-125I added to serum wasconfirmed. However, after zone

elec-trophoresis, ILA was detected in the gamma and beta globulin fractions, whereas no activity was found in the

_alpha,

globulin-albumin fraction. Bolinger, van der Geld, and Willebrands, using

the rat hemidiaphragm bioassay, have reported

similar results (14). In contrast, by means of the more sensitive adipose tissue bioassay, ILA was found in gamma-beta globulin andalpha1 globulin-albumin fractions, yet only the

_alpha,

globulin-albumin migrating ILA has been shown to increase after glucose loading with no change in the level of gamma-beta migrating ILA (17, 19). Thus, serum immunoreactive insulin (IRI), endogen-ously released insulin in response to glucose and serum ILA migrating in the

_alpha,

globulin re-gion are presumably identical. Since the fasting serum level of IRI in normal subjects is under 50 uU/ml (4), which is below the limit of sensi-tivity of the rat hemidiaphragm bioassay, it is not surprising that significant levels of ILA were not detected in the

_alpha,

globulin-albumin fraction. By means of zone electrophoresis using paper (15), cellulose (12, 13), potato starch (14), and Pevikon (17-19), serum ILA has been shown

to migrate in the

ganmma-beta

globulin region, however, its nature has been controversial.

Sev-eral investigators have suggested that the ILA, in gamma-beta globulin preparations is, in fact, due

to insulin bound to these proteins (13. 18, 20). Berson and Yalow have extensively studied this aspect and find no evidence for in vivo or in vitro binding of insulin to serum protein in noninsulin-treated subjects (21), and in the present experi-ments the failure of insulin antiserum to suppress the ILA of gamma and beta globulins appears to support their conclusions. However, partial-to-complete suppression of this ILA has been ob-served by others (13, 18). These discrepancies are not easily reconciled. If insulin is bound to gamma and beta-migrating serum proteins, it seems unusual that it is dissociated at pH 7.4 in

0.15 M phosphate buffer, as was observed during gel filtration of these proteins. Furthermore, the present observations suggest that the molecular weight of the active component of these electro-phoretic preparations approximates40,000-50,000,

whereas both crystalline insulin and serum IRI appeared in a different fraction of lower molecular size.

Inthese studies, the only physicochemical prop-erty common to insulin and this serum insulin-like substance was solubility in acid-ethanol. Previ-ously, serum ILA extracted into acid-ethanol has been attributed to endogenous insulin (22-24);

concept is nolonger tenable. Acid-ethanol extracts of serumdocontain insulin atconcentrations simi-lar to those detected in serum by immunoassay (28), but bioassay of these extracts indicate that the major portion of ILA is unreactive with in-sulin specific antiserum, a finding which in con-junction with its behavior during electrophoresis and gel filtration supports the conclusion that this molecule is not crystalline insulin. Because more ILA is detected in serum after acid-ethanol ex-traction, it has been suggested that the ILA is either "released" or modified by this procedure,

so that it is readily measured by in vitro bioassays (22, 23, 64). However, at present, unequivocal data are not available to support this assumption. In the experiments presented in this paper, the acid-ethanol-extracted ILA, unreactive with in-sulin antibody, behaved identical to that found in unaltered serum as shown by electrophoretic mo-bility and gel filtration.

Ion exchange chromatographic techniques have not been extensively employed in the study of in-sulin or inin-sulin-like activity in serum. Power, Reyes-Leal, and Conn have shown that the ILA in serum globulin fractions obtained by ammonium sulfate precipitation can be eluted from DEAE-cellulose under conditions of decreasing pH and increasing ionic strength of the buffering medium

(65). It has been assumed that this ILA

repre-sents insulin. The present studies of the behavior of ILA in both serum and acid-ethanol extracts of serum demonstrate that, under conditions of stepwise elution, ILA can be separated from beta and gamma globulins but notfrom albumin. That this ILA isnotequatable with insulin is supported l)y the inability of insulin specific antisera to

neu-tralize its biological activity and the distinctly dif-ferent chromatographc properties of both

insulin-131I and crystalline insulin. Although ILA could notbeseparated from albumin onDEAE-cellulose, the fact that the ILA in both serum and acid-ethanol extracts of serum migrates on zone

elec-trophoresis distinct from albumin indicates that

ILA is nota property of thealbumin molecule. The present investigations of the distribution of ILA on the cation exchange resin Dowex 50 confirm the results of Antoniades and coworkers (32-34, 38). Dowex 50 X- 8, 25-50 mesh size,

was originally used as described by Antoniades (60), but was displaced infavor of 100-200 mesh

resin because of apparent incomplete adsorption of ILA. In these studies, aresin to serum ratio of 1: 1 was found to result in maximal retention of ILA. A previous report onthe behavior of crystal-line insulin under various conditions of chroma-tography on Dowex emphasizes the importance

of pH (61). In phosphate buffers at pH below

6.8, insulin is adsorbed to the resin, which

prob-ably reflects diminished solubility below

_pH,

7.0 (66) asinsulin is cationiconly below pH 5.3(66). Since serumapproximates pH 8 when exposed to a normal laboratory environment, immunoassay-able insulin would be expected to pass through the resin and this has been confirmed for both un-labelled and radioun-labelled insulin (33, 61). The failure to find immunoreactive insulin in the resin-adsorbed protein fraction (34, 38) or to in-hibit the ILA of this fraction with insulin

antise-rum (38) (both of which are confirmed by the present studies) conforms to the expected anionic

behavior of insulin at this pH. Because of the ad-sorption of ILA on the resin under conditions

which distinguish it from insulin, Antoniades and

colleagues have referred to this ILA as "bound insulin." It was originally proposed that insulin was bound to a basic protein that imparted a net

cationic charge to the molecular complex;

how-ever, since attempts todissociate immunoassayable insulin from the resin-adsorbed insulin-like

sub-stance have been unsuccessful for the most part (36-38), it has been suggested that "bound

in-sulin" may represent either a combination of

in-sulin molecules associated with other serum

mac-romolecules (38) or possibly a substance

struc-turally distinct from insulin, the physiological significance of which is not understood at

pres-ent (28-31). From our results, it is difficult to make definite conclusions regarding the ionic

na-ture of the insulin-like substances in human se-rum since studies of the fundamental mechanism of adsorption of proteins and other macromole-cules onto ionexchange surfaces arelacking (67). Most certainly, factors in addition to molecular

charge play significant roles in determining the

chromatographic behavior of complex molecules.

This would seemparticularly thecase with serum

ILA since it would appear to behave as a cation on Dowex 50; yet on DEAE-cellulose. it does

not exhibit the characteristics ofa cation.

Although the present studies corroborate the

findings of Antoniades and coworkers that ILA isadsorbed onto Dowex, a perplexing discrepancy exists in the results obtained for the biological activity of this ILA on muscle. Based on data de-rived from the rat hemidiaphragm bioassay of serum and "bound insulin" preparations, with and without the presence of an extract of adipose tis-sue (ATE), it has been concluded that "bound insulin" is inactive on muscle unless dissociated or activated by a low molecular substance pres-ent in adipose tissue extracts (60, 68). In our hands, however, the hemidiaphragm bioassay of the resin-adsorbed protein ("bound insulin") consistently yielded considerable biological ac-tivity in the absence of ATE. Certainly, variations in the method of chromatography and in the bio-assay procedure may account for this difference in part; since comparative studies with and with-out ATE were not done, this question remains unanswered. It should be noted, however, that the specific activities of several eluate fractions bioas-sayed without ATE moderately exceeded values recently reported by others employing ATE (34, 41, 60). Fractionated human serum ILA pre-pared by a variety of other methods was also biologically active on muscle in vitro (38). In addition, "bound insulin" preparations injected

in vivo appear to promote glycogen synthesis in muscle (34). It appears, therefore, that the con-cept of biological inactivity of this insulin-like substance on muscle must be modified and the

problems of hemidiaphragm bioasay of undiluted serum closely examined.

In most of the studies of serum ILA, the

parameters of glucose uptake in isolated muscle

or glucose oxidation in isolated adipose tissue have been used. Thespecificity of theseeffectshas been questioned because glucose utilization in

isolated tissues has been stimulated by environ-mental alterations that include

hyperosmolarity

(69) and anaerobiasis (70), as well as uncoup-ling agents (70) and certain amino acids and hormones (3). Inthe present experiments, in ad-dition to stimulation of glucose uptake, the ILA in the gamma-beta migrating zone from serum,

"albumin,"'

supernatant material, and alkaline eluate protein enhanced glycogen synthesis in the

rat hemidiaphragm, an effect which has not been observed with the above environmental factors and which has been specifically attributed to insulin

(71, 72). While some investigators have demon-strated both stimulation of glucose uptake and gly-cogen synthesis in muscle by chymotrypsin and trypsin (73), others have failed to confirm this finding (74). That chymotrypsin or trypsin can be equated with serum ILA seems unlikely, since the concentration of chymotryptic activity in se-rum is below that required for an in vitro effect

(75,

76). Moreover, the molecular weights of pancreatic preparations of chymotrypsin (25,000 mol wt) and trypsin (20,000 mol wt) (77) differ from that observed for ILA.

Previous studies of some of the physicochemi-cal properties of "atypical" insulin, "bound" in-sulin, and NSILA are in general agreement.

Samaan, Fraser, and Dempster, by dialysis, have estimated the molecular size of "atypical" insulin at greater than 30,000 (9). "Bound" insulin has been reported to migrate in the gamma and beta

globulin region on zone electrophoresis and from its behavior on Sephadex G-100 gel filtration at pH8.0, Antoniades andcolleagues have concluded

that "bound" insulin has a molecular weight

be-tween 60,000 and 100,000 (38).

Bfirgi

and as-sociates have reported beta and

_alpha,

globulin electrophoretic migration of a highly purified

preparation of NSILA (30). On the basis of

Sephadex G-200 filtration of pH

7.2,

the latter workers have also estimated the molecular weight

of the NSILA of serum to range from

70,000

to

150,000.However, whenahighly

purified

prepara-tion of serum NSILAwas

subjected

to

gel

filtra-tion through

G-75,

a broad indefinite

peak

of ILA was obtained, which suggests the

possibility

of multiple

polymeric

forms of this insulin-like

substance. Furthermore, the

highly

purified

se-rum NSILA

appeared

to "dissociate" into

com-pounds with a molecular

weight

approximating

6000 to 10,000 when

subjected

to

gel

filtration in 5 M acetic acid

(30).

In the present

studies,

in

which the unfractionated serum, gamma and beta

globulin fractions from zone

electrophoresis,

acid-ethanol extracts, and alkaline eluate

protein

were

subjected

to gel

filtration,

the major

peak

of

_ILA,

not

suppressed

byinsulin

anitiserum,

waseluted in

avolume

corresponding

to the behavior of a

sub-stance of molecular

weight

of

approximately

40,-000-50,000.

Although

precise

resolution was not

obtained by definition of the volumes

pooled

for

bioassay, the

general

symmetry of distribution of

ILA suggests that the filtration pattern repre-sented a single peak of ILA with a molecular size approximating that of ovalbumin. It is conceivable that the discrepancy in the apparent molecular weight for ILA in neutral buffers between these studies and those of Burgi and coworkers (30) and Antoniades and colleagues (38) may be due

to the improved resolution provided by the in-creased dimensions of the column and the choice of Sephadex G-100 in these experiments.

The importance of pH in "activating" alkaline eluate ILA has previously been reported (33, 78). Apparently, isoelectric precipitation of protein in the alkaline eluate was observed after pH ad-justment to 9.8 to 10. The supernatant resulting from this procedure contained biological activity

on the rat hemidiaphragm without ATE. In the present studies, pH adjustment of the Dowex al-kaline eluate over a wide range (from pH 2 to 12) did not result in observable precipitation of protein nor increase in biological activity at pH 10. Inactivation of biological activity at pH 12, how-ever, did provide correlation with the reports of alkaline instability of a partially purified prepara-tion of serum nonsuppressible ILA (25). In ad-dition, this preparation of ILA was heat stable at 700 and 80'C (30), a finding in agreement with the observed heat stability for Dowex-adsorbed ILA in the present studies.

On the basis of in vitro biological activity, lack of suppression of activity by insulin antiserum, behavior on anion and cation exchange resin, electrophoretic mobility, and molecular weight, it seems probable that "bound insulin," "total

se-rum insulin activity," "atypical insulin" and "non-suppressible insulin-like activity" are identical. Because the physical properties of insulin and ILA

are markedly different, a structural resemblance

to insulin remains conjectural. If this ILA con-sists of insulin complexed to a macromolecule in

serum as has been suggested (33), it is neces-sary to invoke astrong covalent bond between in-sulinand its "carrier," since the procedure of acid-ethanol extraction would be expected to dissoci-ate molecules aggregated by weaker forces such as hydrogen bonding or Van der Waal's forces.

Be-cause there is as yet no convincing evidence that

ILA is related structurally to insulin, the term

"nonsuppressible insulin-like activity" (NSILA),

proposed by Froesch, Burgi, Ramseier, Bally, and

Labhart (29), would seem to us to be the most

appropriate description of this substance.

Thus, from the studies reported herein, it seems

justified to conclude that both insulin and a single molecular species distinct from insulin by

physi-cochemcial criteria contribute to the measurement of ILA in in vitro bioassay systems.

ACKNOWLEDGMENTS

We wish to thank Mrs. Gail Movius and Mrs. Agnes

Lee for their assistance.

These studies were supported by U. S. Public Health Service grants 1-RO1-AM-1215-01, AM-02456, TI AM-5020-12, and Postdoctoral Research Fellowship 1-F2-AM-30, 777-01 from the Institute of Arthritis and

Meta-bolic Diseases. Additional support was provided by an

Institutional Research Grant from the American Cancer Society, Boeing Good Neighbors Fund, and Fund 171

from the University of Washington.

REFERENCES

1. Vallance-Owen, J., and P. H. Wright. 1960. Assay of insulin in blood. Physiol. Rev. 40: 219.

2. Martin, D. B., A. E. Renold, and Y. M. Dagenais.

1958. An assay for insulin-like activity using rat adi-pose tissue. Lancet. 2: 76.

3. Randle, P. J. 1957. Insulin in blood. Ciba Found. Colloq. Endocrinol. 11: 115.

4. Berson, S. A., and R. S. Yalow. 1965. Some cur-rent controversies in diabetes research. Diabetes. 14: 549.

5. Rasio, E. A., J. S. Soeldner, and G. F. Cahill, Jr. 1965. Insulin and insulin-like activity in serum and

extravascular fluid. Diabetologia. 1: 125.

6. Steinke, J., A. Sirek, V. Lauris, F. D. W. Lukens,

and A. E. Renold. 1962. Measurement of small quan-tities of insulin-like activity with rat adipose tissue.

III. Persistence of serum insulin-like activity after pancreatectomy. J. Clin. Invest. 41: 1699.

7. Egdahl, R. H., and H. Goldberg. 1962. Pancreatic and hepatic influences on serum insulin-like activity

in the dog. Surg. Gynecol. Obstet. 114: 202.

8. Schoeffling, K., H. Ditschuneit, R. Petzoldt, J. Beyer,

E. F. Pfeiffer, A. Sirek, H. Geerling, and 0. Sirek.

1965. Serum insulin-likeactivityinhypophysectomized

anddepancreatized (Houssay) dogs.Diabetes. 14:658. 9. Samaan, N., R. Fraser, and W. J. Dempster. 1963. The "typical" and "atypical" forms of serum insulin. Diabetes. 12: 339.

10. Antoniades, H. N. 1965. Extrapancreatic regulation

of insulin activity in human beings. In On the Na-ture and Treatment of Diabetes. B. S. Leibel and G. A. Wrenshall, editors. ExcerptaMed. 84: 194. 11. Beigelman, P. M., H. N. Antoniades, F. C. Goetz,

A. E. Renold, J. L. Oncley, and G. W. Thorn.

1956. Insulin-like activity of human plasma